apunkt/plano
1
0
Fork 0
mirror of https://github.com/katanemo/plano.git synced 2026-07-02 15:51:02 +02:00
Code Issues Projects Releases Packages Wiki Activity

more refactoring to clean code and make stream_context.rs work

Browse source
This commit is contained in:
Salman Paracha 2025-08-10 23:10:28 -07:00
parent d4ca70d177
commit df3aa17d67
23 changed files with 545 additions and 1321 deletions
Download patch file Download diff file Expand all files Collapse all files

19
crates/common/src/configuration.rs
View file

@ -178,13 +178,10 @@ impl Display for LlmProviderType {
} }
impl LlmProviderType { impl LlmProviderType {
/// Create a Provider from this LlmProviderType /// Get the ProviderId for this LlmProviderType
/// This is the main method for stream_context to get provider-specific interfaces /// Used with the new function-based hermesllm API
pub fn create_provider(&self) -> hermesllm::Provider { pub fn to_provider_id(&self) -> hermesllm::ProviderId {
use hermesllm::{ProviderId, Provider}; hermesllm::ProviderId::from(self.to_string().as_str())
let provider_id = ProviderId::from(self.to_string().as_str());
Provider::new(provider_id)
} }
} }
@ -264,10 +261,10 @@ impl Display for LlmProvider {
} }
impl LlmProvider { impl LlmProvider {
/// Create a Provider for this LlmProvider /// Get the ProviderId for this LlmProvider
/// This is a convenience method that delegates to the provider_interface /// Used with the new function-based hermesllm API
pub fn create_provider(&self) -> hermesllm::Provider { pub fn to_provider_id(&self) -> hermesllm::ProviderId {
self.provider_interface.create_provider() self.provider_interface.to_provider_id()
} }
} }

8
crates/hermesllm/README.md
View file

@ -35,15 +35,17 @@ _Replace the path with the appropriate location if using as a workspace member o
Construct a chat completion request using the builder pattern: Construct a chat completion request using the builder pattern:
```rust ```rust
use hermesllm::Provider; use hermesllm::{create_provider, ProviderId};
use hermesllm::providers::openai::types::ChatCompletionsRequest; use hermesllm::providers::openai::types::ChatCompletionsRequest;
let request = ChatCompletionsRequest::builder("gpt-3.5-turbo", vec![Message::new("Hi".to_string())]) let request = ChatCompletionsRequest::builder("gpt-3.5-turbo", vec![Message::new("Hi".to_string())])
.build() .build()
.expect("Failed to build OpenAIRequest"); .expect("Failed to build OpenAIRequest");
// Convert to bytes for a specific provider // Create a provider and convert request to bytes
let bytes = request.to_bytes(Provider::OpenAI)?; let provider = create_provider(ProviderId::OpenAI);
let bytes = serde_json::to_vec(&request)?;
let parsed_request = provider.try_request_from_bytes(&bytes)?;
``` ```
## API Overview ## API Overview

207
crates/hermesllm/src/apis/openai.rs
View file

@ -3,6 +3,7 @@ use serde_json::Value;
use serde_with::skip_serializing_none; use serde_with::skip_serializing_none;
use std::collections::HashMap; use std::collections::HashMap;
use crate::{providers::ProviderRequestError, ConversionMode, ProviderRequest};
use super::ApiDefinition; use super::ApiDefinition;
// ============================================================================ // ============================================================================
@ -424,6 +425,212 @@ pub struct StreamOptions {
pub include_usage: Option<bool>, pub include_usage: Option<bool>,
} }
/// ============================================================================
/// OpenAI Provider Request Wrapper
/// ============================================================================
impl ProviderRequest for ChatCompletionsRequest {
fn model(&self) -> &str {
&self.model
}
fn is_streaming(&self) -> bool {
self.stream.unwrap_or_default()
}
fn set_streaming_options(&mut self) {
self.stream = Some(true);
if self.stream_options.is_none() {
self.stream_options = Some(StreamOptions { include_usage: Some(true) });
}
}
fn extract_messages_text(&self) -> String {
self.messages.iter().fold(String::new(), |acc, m| {
acc + " " + &match &m.content {
MessageContent::Text(text) => text.clone(),
MessageContent::Parts(parts) => parts.iter().map(|part| match part {
ContentPart::Text { text } => text.clone(),
ContentPart::ImageUrl { .. } => "[Image]".to_string(),
}).collect::<Vec<_>>().join(" ")
}
})
}
fn extract_user_message(&self) -> Option<String> {
self.messages.last().and_then(|msg| {
match &msg.content {
MessageContent::Text(text) => Some(text.clone()),
MessageContent::Parts(_) => None, // No user message in parts
}
})
}
fn to_provider_bytes(&self, mode: ConversionMode) -> Result<Vec<u8>, ProviderRequestError> {
match mode {
ConversionMode::Compatible | ConversionMode::Passthrough => {
serde_json::to_vec(&self).map_err(|e| ProviderRequestError {
message: format!("Failed to serialize OpenAI request: {}", e),
source: Some(Box::new(e)),
})
}
}
}
}
// ============================================================================
// STREAMING SUPPORT
// ============================================================================
use crate::providers::traits::{ProviderResponse, ProviderStreamResponse, ProviderStreamResponseIter, TokenUsage};
// Direct implementation of ProviderResponse on ChatCompletionsResponse
impl ProviderResponse for ChatCompletionsResponse {
fn usage(&self) -> Option<&dyn TokenUsage> {
Some(&self.usage)
}
}
// Implementation of TokenUsage for OpenAI Usage type
impl TokenUsage for Usage {
fn completion_tokens(&self) -> usize {
self.completion_tokens as usize
}
fn prompt_tokens(&self) -> usize {
self.prompt_tokens as usize
}
fn total_tokens(&self) -> usize {
self.total_tokens as usize
}
}
// Direct implementation of ProviderStreamResponse trait on ChatCompletionsStreamResponse
impl ProviderStreamResponse for ChatCompletionsStreamResponse {
fn content_delta(&self) -> Option<&str> {
self.choices
.first()
.and_then(|choice| choice.delta.content.as_deref())
}
fn is_final(&self) -> bool {
self.choices
.first()
.map(|choice| choice.finish_reason.is_some())
.unwrap_or(false)
}
fn role(&self) -> Option<&str> {
self.choices
.first()
.and_then(|choice| choice.delta.role.as_ref().map(|r| match r {
Role::System => "system",
Role::User => "user",
Role::Assistant => "assistant",
Role::Tool => "tool",
}))
}
}
// Error type for streaming operations
#[derive(Debug, thiserror::Error)]
pub enum OpenAIStreamError {
#[error("JSON parsing error: {0}")]
JsonError(#[from] serde_json::Error),
#[error("UTF-8 parsing error: {0}")]
Utf8Error(#[from] std::str::Utf8Error),
#[error("Invalid streaming data: {0}")]
InvalidStreamingData(String),
}
/// SSE-based streaming iterator for OpenAI chat completions
/// Implements ProviderStreamResponseIter directly
pub struct SseChatCompletionIter<I>
where
I: Iterator,
I::Item: AsRef<str>,
{
lines: I,
}
impl<I> SseChatCompletionIter<I>
where
I: Iterator,
I::Item: AsRef<str>,
{
pub fn new(lines: I) -> Self {
Self { lines }
}
}
impl<I> Iterator for SseChatCompletionIter<I>
where
I: Iterator,
I::Item: AsRef<str>,
{
type Item = Result<Box<dyn ProviderStreamResponse>, Box<dyn std::error::Error + Send + Sync>>;
fn next(&mut self) -> Option<Self::Item> {
for line in &mut self.lines {
let line = line.as_ref();
if line.is_empty() {
continue;
}
if line.starts_with("data: ") {
let data = &line[6..]; // Remove "data: " prefix
if data == "[DONE]" {
return None;
}
if data == r#"{"type": "ping"}"# {
continue; // Skip ping messages - that is usually from anthropic
}
match serde_json::from_str::<ChatCompletionsStreamResponse>(data) {
Ok(response) => return Some(Ok(Box::new(response))),
Err(e) => return Some(Err(Box::new(
OpenAIStreamError::InvalidStreamingData(format!("Error parsing: {}, data: {}", e, data))
))),
}
}
}
None
}
}
impl<I> ProviderStreamResponseIter for SseChatCompletionIter<I>
where
I: Iterator + Send + Sync,
I::Item: AsRef<str>,
{
// Just marking that this type implements the trait - no additional methods needed
}
// ============================================================================
// PARAMETERIZED CONVERSIONS FOR PROVIDER FUNCTIONS
// ============================================================================
use crate::providers::ProviderId;
/// Parameterized conversion for ChatCompletionsRequest
impl TryFrom<(&[u8], &ProviderId)> for ChatCompletionsRequest {
type Error = OpenAIStreamError;
fn try_from((bytes, _provider_id): (&[u8], &ProviderId)) -> Result<Self, Self::Error> {
serde_json::from_slice(bytes).map_err(OpenAIStreamError::from)
}
}
/// Parameterized conversion for ChatCompletionsResponse
impl TryFrom<(&[u8], &ProviderId)> for ChatCompletionsResponse {
type Error = OpenAIStreamError;
fn try_from((bytes, _provider_id): (&[u8], &ProviderId)) -> Result<Self, Self::Error> {
serde_json::from_slice(bytes).map_err(OpenAIStreamError::from)
}
}
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use super::*; use super::*;

118
crates/hermesllm/src/lib.rs
View file

@ -7,10 +7,11 @@ pub mod clients;
// Re-export important types and traits // Re-export important types and traits
pub use providers::{ pub use providers::{
ProviderId, Provider, ConversionMode, ProviderId, ConversionMode,
ProviderInterface, ProviderRequest, ProviderResponse, ProviderRequest, ProviderResponse, ProviderStreamResponse, ProviderStreamResponseIter,
TokenUsage, StreamChunk, StreamingResponse, TokenUsage,
OpenAIProvider try_request_from_bytes, try_response_from_bytes, try_streaming_from_bytes,
has_compatible_api, supported_apis
}; };
#[cfg(test)] #[cfg(test)]
@ -26,70 +27,71 @@ mod tests {
} }
#[test] #[test]
fn test_provider_api_paths() { fn test_provider_api_compatibility() {
assert_eq!(ProviderId::OpenAI.api_path(), "/v1/chat/completions"); assert!(has_compatible_api(&ProviderId::OpenAI, "/v1/chat/completions"));
assert_eq!(ProviderId::Groq.api_path(), "/openai/v1/chat/completions"); assert!(!has_compatible_api(&ProviderId::OpenAI, "/v1/embeddings"));
assert_eq!(ProviderId::Mistral.api_path(), "/v1/chat/completions");
assert_eq!(ProviderId::Arch.api_path(), "/v1/chat/completions");
}
#[test]
fn test_provider_openai_format_support() {
assert!(ProviderId::OpenAI.supports_openai_format());
assert!(ProviderId::Groq.supports_openai_format());
assert!(ProviderId::Mistral.supports_openai_format());
assert!(ProviderId::Arch.supports_openai_format());
assert!(!ProviderId::Gemini.supports_openai_format());
assert!(!ProviderId::Claude.supports_openai_format());
}
#[test]
fn test_provider_instance_creation() {
let provider = Provider::new(ProviderId::OpenAI);
assert!(provider.has_compatible_api("/v1/chat/completions"));
assert!(!provider.has_compatible_api("/v1/embeddings"));
} }
#[test] #[test]
fn test_provider_supported_apis() { fn test_provider_supported_apis() {
let provider = Provider::new(ProviderId::OpenAI); let apis = supported_apis(&ProviderId::OpenAI);
assert!(apis.contains(&"/v1/chat/completions"));
let supported_apis = provider.supported_apis();
assert!(supported_apis.contains(&"/v1/chat/completions"));
// Test that provider supports the expected API endpoints // Test that provider supports the expected API endpoints
assert!(provider.has_compatible_api("/v1/chat/completions")); assert!(has_compatible_api(&ProviderId::OpenAI, "/v1/chat/completions"));
} }
#[test] #[test]
fn test_provider_extract_user_message() { fn test_provider_request_parsing() {
use crate::apis::openai::{ChatCompletionsRequest, Message, MessageContent}; // Test with a sample JSON request
let json_request = r#"{
let provider = Provider::new(ProviderId::OpenAI); "model": "gpt-4",
"messages": [
// Test with text message {
let request = ChatCompletionsRequest { "role": "system",
model: "gpt-4".to_string(), "content": "You are a helpful assistant"
messages: vec![
Message {
role: crate::apis::openai::Role::System,
content: MessageContent::Text("You are a helpful assistant".to_string()),
name: None,
tool_calls: None,
tool_call_id: None,
}, },
Message { {
role: crate::apis::openai::Role::User, "role": "user",
content: MessageContent::Text("Hello, world!".to_string()), "content": "Hello!"
name: None, }
tool_calls: None, ]
tool_call_id: None, }"#;
},
],
..Default::default()
};
let user_message = provider.extract_user_message(&request); let result = try_request_from_bytes(json_request.as_bytes(), &ProviderId::OpenAI);
assert_eq!(user_message, Some("Hello, world!".to_string())); assert!(result.is_ok());
let request = result.unwrap();
assert_eq!(request.model(), "gpt-4");
assert_eq!(request.extract_user_message(), Some("Hello!".to_string()));
}
#[test]
fn test_provider_streaming_response() {
// Test streaming response parsing with sample SSE data
let sse_data = r#"data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"gpt-4","choices":[{"index":0,"delta":{"role":"assistant","content":"Hello"},"finish_reason":null}]}
data: [DONE]
"#;
let result = try_streaming_from_bytes(sse_data.as_bytes(), &ProviderId::OpenAI, ConversionMode::Passthrough);
assert!(result.is_ok());
let mut streaming_response = result.unwrap();
// Test that we can iterate over chunks - it's just an iterator now!
let first_chunk = streaming_response.next();
assert!(first_chunk.is_some());
let chunk_result = first_chunk.unwrap();
assert!(chunk_result.is_ok());
let chunk = chunk_result.unwrap();
assert_eq!(chunk.content_delta(), Some("Hello"));
assert!(!chunk.is_final());
// Test that stream ends properly
let final_chunk = streaming_response.next();
assert!(final_chunk.is_none());
} }
} }

6
crates/hermesllm/src/providers/arch/mod.rs
View file

@ -1,6 +0,0 @@
//! Arch provider implementation
//!
//! Arch uses OpenAI-compatible API format
pub mod provider;
pub use provider::ArchProvider;

88
crates/hermesllm/src/providers/arch/provider.rs
View file

@ -1,88 +0,0 @@
//! Arch provider implementation
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Arch provider implementation
#[derive(Debug, Clone)]
pub struct ArchProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for ArchProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for ArchProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for ArchProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for ArchProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
matches!(api_path, "/v1/chat/completions")
}
fn supported_apis(&self) -> Vec<&'static str> {
vec!["/v1/chat/completions"]
}
}

7
crates/hermesllm/src/providers/claude/mod.rs
View file

@ -1,7 +0,0 @@
//! Claude provider implementation
//!
//! Claude will use a different API format in the future (/v1/messages)
//! For now, fallback to OpenAI-compatible format
pub mod provider;
pub use provider::ClaudeProvider;

93
crates/hermesllm/src/providers/claude/provider.rs
View file

@ -1,93 +0,0 @@
//! Claude provider implementation
//!
//! TODO: Implement Claude-specific API format (/v1/messages) when needed
//! For now, uses OpenAI-compatible format as fallback
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Claude provider implementation
#[derive(Debug, Clone)]
pub struct ClaudeProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for ClaudeProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for ClaudeProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for ClaudeProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for ClaudeProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
// TODO: Update when Claude API is fully implemented
matches!(api_path, "/v1/chat/completions" | "/v1/messages")
}
fn supported_apis(&self) -> Vec<&'static str> {
// TODO: Update when Claude API is fully implemented
vec!["/v1/messages"]
}
}

6
crates/hermesllm/src/providers/deepseek/mod.rs
View file

@ -1,6 +0,0 @@
//! Deepseek provider implementation
//!
//! Deepseek uses OpenAI-compatible API format
pub mod provider;
pub use provider::DeepseekProvider;

88
crates/hermesllm/src/providers/deepseek/provider.rs
View file

@ -1,88 +0,0 @@
//! Deepseek provider implementation
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Deepseek provider implementation
#[derive(Debug, Clone)]
pub struct DeepseekProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for DeepseekProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for DeepseekProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for DeepseekProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for DeepseekProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
matches!(api_path, "/v1/chat/completions")
}
fn supported_apis(&self) -> Vec<&'static str> {
vec!["/v1/chat/completions"]
}
}

7
crates/hermesllm/src/providers/gemini/mod.rs
View file

@ -1,7 +0,0 @@
//! Gemini provider implementation
//!
//! Gemini will use a different API format in the future
//! For now, fallback to OpenAI-compatible format
pub mod provider;
pub use provider::GeminiProvider;

93
crates/hermesllm/src/providers/gemini/provider.rs
View file

@ -1,93 +0,0 @@
//! Gemini provider implementation
//!
//! This module contains the Gemini provider that handles Google's Gemini API format
//! requests in OpenAI-compatible format.
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Gemini provider implementation
#[derive(Debug, Clone)]
pub struct GeminiProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for GeminiProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for GeminiProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for GeminiProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for GeminiProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
// TODO: Update when Gemini API is fully implemented
matches!(api_path, "/v1/chat/completions" | "/v1/models")
}
fn supported_apis(&self) -> Vec<&'static str> {
// TODO: Update when Gemini API is fully implemented
vec!["/v1/models"]
}
}

7
crates/hermesllm/src/providers/github/mod.rs
View file

@ -1,7 +0,0 @@
//! GitHub provider implementation
//!
//! GitHub will use a different API format in the future (/models)
//! For now, fallback to OpenAI-compatible format
pub mod provider;
pub use provider::GitHubProvider;

93
crates/hermesllm/src/providers/github/provider.rs
View file

@ -1,93 +0,0 @@
//! GitHub provider implementation
//!
//! This module contains the GitHub provider that handles GitHub API format
//! requests in OpenAI-compatible format.
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// GitHub provider implementation
#[derive(Debug, Clone)]
pub struct GitHubProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for GitHubProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for GitHubProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for GitHubProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for GitHubProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
// TODO: Update when GitHub API is fully implemented
matches!(api_path, "/v1/chat/completions" | "/models")
}
fn supported_apis(&self) -> Vec<&'static str> {
// TODO: Update when GitHub API is fully implemented
vec!["/models"]
}
}

6
crates/hermesllm/src/providers/groq/mod.rs
View file

@ -1,6 +0,0 @@
//! Groq provider implementation
//!
//! Groq uses OpenAI-compatible API format but with different endpoints
pub mod provider;
pub use provider::GroqProvider;

91
crates/hermesllm/src/providers/groq/provider.rs
View file

@ -1,91 +0,0 @@
//! Groq provider implementation
//!
//! This module contains the Groq provider that handles Groq API format requests.
//! Groq uses OpenAI-compatible format but may have provider-specific nuances.
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Groq provider implementation
#[derive(Debug, Clone)]
pub struct GroqProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for GroqProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for GroqProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for GroqProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for GroqProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
matches!(api_path, "/v1/chat/completions" | "/openai/v1/chat/completions")
}
fn supported_apis(&self) -> Vec<&'static str> {
vec!["/openai/v1/chat/completions"]
}
}

6
crates/hermesllm/src/providers/mistral/mod.rs
View file

@ -1,6 +0,0 @@
//! Mistral provider implementation
//!
//! Mistral uses OpenAI-compatible API format
pub mod provider;
pub use provider::MistralProvider;

88
crates/hermesllm/src/providers/mistral/provider.rs
View file

@ -1,88 +0,0 @@
//! Mistral provider implementation
use crate::providers::{ProviderInterface, ConversionMode};
use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, Usage};
use crate::providers::traits::{ProviderRequest, ProviderResponse, StreamingResponse};
use crate::providers::openai::provider::{OpenAIProvider, OpenAIStreamingResponse, OpenAIApiError};
/// Mistral provider implementation
#[derive(Debug, Clone)]
pub struct MistralProvider;
// Trait implementations that delegate to OpenAI
impl ProviderRequest for MistralProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::try_from_bytes(&openai_provider, bytes)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, provider: super::super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderRequest::to_provider_bytes(&openai_provider, request, provider, mode)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
let openai_provider = OpenAIProvider;
ProviderRequest::set_streaming_options(&openai_provider, request)
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_messages_text(&openai_provider, request)
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
let openai_provider = OpenAIProvider;
ProviderRequest::extract_user_message(&openai_provider, request)
}
}
impl ProviderResponse for MistralProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let openai_provider = OpenAIProvider;
ProviderResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
let openai_provider = OpenAIProvider;
ProviderResponse::extract_usage_counts(&openai_provider, response)
}
}
impl StreamingResponse for MistralProvider {
type Error = OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider: &super::super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let openai_provider = OpenAIProvider;
StreamingResponse::try_from_bytes(&openai_provider, bytes, provider, mode)
}
}
impl ProviderInterface for MistralProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
matches!(api_path, "/v1/chat/completions")
}
fn supported_apis(&self) -> Vec<&'static str> {
vec!["/v1/chat/completions"]
}
}

233
crates/hermesllm/src/providers/mod.rs
View file

@ -5,24 +5,11 @@
pub mod traits; pub mod traits;
pub mod openai; pub mod openai;
pub mod groq;
pub mod mistral;
pub mod deepseek;
pub mod arch;
pub mod gemini;
pub mod claude;
pub mod github;
// Re-export the main interfaces // Re-export the main interfaces
pub use traits::*; pub use traits::*;
pub use openai::OpenAIProvider; // Note: OpenAIProvider has been deprecated in favor of function-based approach
pub use groq::GroqProvider; // OpenAI functionality is accessed through openai::builder and openai::types modules
pub use mistral::MistralProvider;
pub use deepseek::DeepseekProvider;
pub use arch::ArchProvider;
pub use gemini::GeminiProvider;
pub use claude::ClaudeProvider;
pub use github::GitHubProvider;
use std::fmt::Display; use std::fmt::Display;
@ -69,219 +56,3 @@ impl Display for ProviderId {
} }
} }
} }
impl ProviderId {
/// Get the API endpoint path for this provider
pub fn api_path(&self) -> &'static str {
match self {
ProviderId::OpenAI => "/v1/chat/completions",
ProviderId::Groq => "/openai/v1/chat/completions",
ProviderId::Gemini => "/v1/models", // TODO: Update when Gemini API is implemented
ProviderId::Claude => "/v1/messages", // TODO: Update when Claude API is implemented
ProviderId::Mistral => "/v1/chat/completions",
ProviderId::Deepseek => "/v1/chat/completions",
ProviderId::GitHub => "/models", // TODO: Update when GitHub models API is implemented
ProviderId::Arch => "/v1/chat/completions",
}
}
/// Check if this provider supports OpenAI v1/chat/completions API format
pub fn supports_openai_format(&self) -> bool {
matches!(
self,
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek | ProviderId::Arch
)
}
}
/// Enum for dynamic dispatch of provider instances
pub enum Provider {
OpenAI(OpenAIProvider, ProviderId),
Groq(GroqProvider, ProviderId),
Mistral(MistralProvider, ProviderId),
Deepseek(DeepseekProvider, ProviderId),
Arch(ArchProvider, ProviderId),
Gemini(GeminiProvider, ProviderId),
Claude(ClaudeProvider, ProviderId),
GitHub(GitHubProvider, ProviderId),
}
impl Provider {
/// Create a provider instance from a provider ID
pub fn new(id: ProviderId) -> Self {
match id {
ProviderId::OpenAI => Provider::OpenAI(OpenAIProvider, id),
ProviderId::Groq => Provider::Groq(GroqProvider, id),
ProviderId::Mistral => Provider::Mistral(MistralProvider, id),
ProviderId::Deepseek => Provider::Deepseek(DeepseekProvider, id),
ProviderId::Arch => Provider::Arch(ArchProvider, id),
ProviderId::Gemini => Provider::Gemini(GeminiProvider, id),
ProviderId::Claude => Provider::Claude(ClaudeProvider, id),
ProviderId::GitHub => Provider::GitHub(GitHubProvider, id),
}
}
/// Get the provider ID
pub fn id(&self) -> ProviderId {
match self {
Provider::OpenAI(_, id) => *id,
Provider::Groq(_, id) => *id,
Provider::Mistral(_, id) => *id,
Provider::Deepseek(_, id) => *id,
Provider::Arch(_, id) => *id,
Provider::Gemini(_, id) => *id,
Provider::Claude(_, id) => *id,
Provider::GitHub(_, id) => *id,
}
}
}
// Implement traits directly on the Provider enum
impl ProviderRequest for Provider {
type Error = openai::provider::OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<crate::apis::openai::ChatCompletionsRequest, Self::Error> {
match self {
Provider::OpenAI(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Groq(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Mistral(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Deepseek(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Arch(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Gemini(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::Claude(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
Provider::GitHub(provider, _) => ProviderRequest::try_from_bytes(provider, bytes),
}
}
fn to_provider_bytes(&self, request: &crate::apis::openai::ChatCompletionsRequest, provider_id: super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
match self {
Provider::OpenAI(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Groq(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Mistral(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Deepseek(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Arch(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Gemini(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::Claude(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
Provider::GitHub(provider, _) => ProviderRequest::to_provider_bytes(provider, request, provider_id, mode),
}
}
fn extract_model<'a>(&self, request: &'a crate::apis::openai::ChatCompletionsRequest) -> &'a str {
// Since all providers use the same implementation, just use the first one
&request.model
}
fn is_streaming(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> bool {
// Since all providers use the same implementation, just use the first one
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut crate::apis::openai::ChatCompletionsRequest) {
match self {
Provider::OpenAI(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Groq(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Mistral(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Deepseek(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Arch(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Gemini(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::Claude(provider, _) => ProviderRequest::set_streaming_options(provider, request),
Provider::GitHub(provider, _) => ProviderRequest::set_streaming_options(provider, request),
}
}
fn extract_messages_text(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> String {
match self {
Provider::OpenAI(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Groq(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Mistral(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Deepseek(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Arch(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Gemini(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::Claude(provider, _) => ProviderRequest::extract_messages_text(provider, request),
Provider::GitHub(provider, _) => ProviderRequest::extract_messages_text(provider, request),
}
}
fn extract_user_message(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> Option<String> {
match self {
Provider::OpenAI(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Groq(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Mistral(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Deepseek(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Arch(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Gemini(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::Claude(provider, _) => ProviderRequest::extract_user_message(provider, request),
Provider::GitHub(provider, _) => ProviderRequest::extract_user_message(provider, request),
}
}
}
impl ProviderResponse for Provider {
type Error = openai::provider::OpenAIApiError;
type Usage = crate::apis::openai::Usage;
fn try_from_bytes(&self, bytes: &[u8], provider_id: &super::ProviderId, mode: ConversionMode) -> Result<crate::apis::openai::ChatCompletionsResponse, Self::Error> {
match self {
Provider::OpenAI(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Groq(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Mistral(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Deepseek(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Arch(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Gemini(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Claude(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::GitHub(provider, _) => ProviderResponse::try_from_bytes(provider, bytes, provider_id, mode),
}
}
fn usage<'a>(&self, response: &'a crate::apis::openai::ChatCompletionsResponse) -> Option<&'a Self::Usage> {
// Since all providers use the same implementation, just use the direct implementation
Some(&response.usage)
}
}
impl StreamingResponse for Provider {
type Error = openai::provider::OpenAIApiError;
type StreamChunk = crate::apis::openai::ChatCompletionsStreamResponse;
type StreamingIter = openai::provider::OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], provider_id: &super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
match self {
Provider::OpenAI(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Groq(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Mistral(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Deepseek(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Arch(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Gemini(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::Claude(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
Provider::GitHub(provider, _) => StreamingResponse::try_from_bytes(provider, bytes, provider_id, mode),
}
}
}
impl ProviderInterface for Provider {
fn has_compatible_api(&self, api_path: &str) -> bool {
match self {
Provider::OpenAI(provider, _) => provider.has_compatible_api(api_path),
Provider::Groq(provider, _) => provider.has_compatible_api(api_path),
Provider::Mistral(provider, _) => provider.has_compatible_api(api_path),
Provider::Deepseek(provider, _) => provider.has_compatible_api(api_path),
Provider::Arch(provider, _) => provider.has_compatible_api(api_path),
Provider::Gemini(provider, _) => provider.has_compatible_api(api_path),
Provider::Claude(provider, _) => provider.has_compatible_api(api_path),
Provider::GitHub(provider, _) => provider.has_compatible_api(api_path),
}
}
fn supported_apis(&self) -> Vec<&'static str> {
match self {
Provider::OpenAI(provider, _) => provider.supported_apis(),
Provider::Groq(provider, _) => provider.supported_apis(),
Provider::Mistral(provider, _) => provider.supported_apis(),
Provider::Deepseek(provider, _) => provider.supported_apis(),
Provider::Arch(provider, _) => provider.supported_apis(),
Provider::Gemini(provider, _) => provider.supported_apis(),
Provider::Claude(provider, _) => provider.supported_apis(),
Provider::GitHub(provider, _) => provider.supported_apis(),
}
}
}

10
crates/hermesllm/src/providers/openai/mod.rs
View file

@ -1,6 +1,10 @@
pub mod builder; pub mod builder;
pub mod types; pub mod types;
pub mod provider;
// Re-export the main provider // Re-export the main types and builder functionality
pub use provider::OpenAIProvider; pub use crate::apis::openai::{ChatCompletionsRequest, ChatCompletionsResponse, ChatCompletionsStreamResponse};
pub use builder::*;
pub use types::*;
// Note: The OpenAIProvider struct has been deprecated in favor of the function-based approach in traits.rs
// All provider functionality is now accessed through try_request_from_bytes, try_response_from_bytes, etc.

217
crates/hermesllm/src/providers/openai/provider.rs
View file

@ -1,217 +0,0 @@
//! OpenAI provider interface implementations
use crate::apis::openai::*;
use crate::providers::traits::*;
// Simple error type for OpenAI API operations
#[derive(Debug, thiserror::Error)]
pub enum OpenAIApiError {
#[error("JSON parsing error: {0}")]
JsonError(#[from] serde_json::Error),
#[error("UTF-8 parsing error: {0}")]
Utf8Error(#[from] std::str::Utf8Error),
#[error("Invalid streaming data: {0}")]
InvalidStreamingData(String),
#[error("Request conversion error: {0}")]
RequestConversionError(String),
}
// ============================================================================
// OpenAI Provider Definition
// ============================================================================
pub struct OpenAIProvider;
// Create a concrete streaming response type to avoid lifetime issues
pub struct OpenAIStreamingResponse {
lines: Vec<String>,
current_index: usize,
}
impl OpenAIStreamingResponse {
fn new(data: String) -> Self {
let lines: Vec<String> = data.lines().map(|s| s.to_string()).collect();
Self {
lines,
current_index: 0,
}
}
}
impl Iterator for OpenAIStreamingResponse {
type Item = Result<ChatCompletionsStreamResponse, OpenAIApiError>;
fn next(&mut self) -> Option<Self::Item> {
while self.current_index < self.lines.len() {
let line = &self.lines[self.current_index];
self.current_index += 1;
if let Some(data) = line.strip_prefix("data: ") {
let data = data.trim();
if data == "[DONE]" {
return None;
}
if data == r#"{"type": "ping"}"# {
continue; // Skip ping messages
}
return Some(
serde_json::from_str::<ChatCompletionsStreamResponse>(data).map_err(|e| {
OpenAIApiError::InvalidStreamingData(format!("Error parsing: {}, data: {}", e, data))
}),
);
}
}
None
}
}
impl ProviderInterface for OpenAIProvider {
fn has_compatible_api(&self, api_path: &str) -> bool {
api_path == "/v1/chat/completions"
}
fn supported_apis(&self) -> Vec<&'static str> {
vec!["/v1/chat/completions"]
}
}
// Direct trait implementations on OpenAIProvider
impl ProviderRequest for OpenAIProvider {
type Error = OpenAIApiError;
fn try_from_bytes(&self, bytes: &[u8]) -> Result<ChatCompletionsRequest, Self::Error> {
let s = std::str::from_utf8(bytes)?;
Ok(serde_json::from_str(s)?)
}
fn to_provider_bytes(&self, request: &ChatCompletionsRequest, _provider: super::super::ProviderId, _mode: ConversionMode) -> Result<Vec<u8>, Self::Error> {
Ok(serde_json::to_vec(request)?)
}
fn extract_model<'a>(&self, request: &'a ChatCompletionsRequest) -> &'a str {
&request.model
}
fn is_streaming(&self, request: &ChatCompletionsRequest) -> bool {
request.stream.unwrap_or_default()
}
fn set_streaming_options(&self, request: &mut ChatCompletionsRequest) {
if request.stream_options.is_none() {
request.stream_options = Some(StreamOptions {
include_usage: Some(true),
});
}
}
fn extract_messages_text(&self, request: &ChatCompletionsRequest) -> String {
request.messages
.iter()
.fold(String::new(), |acc, m| {
acc + " " + &match &m.content {
MessageContent::Text(text) => text.clone(),
MessageContent::Parts(parts) => {
parts.iter().map(|part| match part {
ContentPart::Text { text } => text.clone(),
ContentPart::ImageUrl { .. } => "[Image]".to_string(),
}).collect::<Vec<_>>().join(" ")
}
}
})
}
fn extract_user_message(&self, request: &ChatCompletionsRequest) -> Option<String> {
request.messages.last().and_then(|msg| {
match &msg.content {
MessageContent::Text(text) => Some(text.clone()),
MessageContent::Parts(parts) => {
// Extract text from content parts, ignoring images
let text_parts: Vec<String> = parts
.iter()
.filter_map(|part| match part {
ContentPart::Text { text } => Some(text.clone()),
ContentPart::ImageUrl { .. } => None,
})
.collect();
if text_parts.is_empty() {
None
} else {
Some(text_parts.join(" "))
}
}
}
})
}
}
impl ProviderResponse for OpenAIProvider {
type Error = OpenAIApiError;
type Usage = Usage;
fn try_from_bytes(&self, bytes: &[u8], _provider: &super::super::ProviderId, _mode: ConversionMode) -> Result<ChatCompletionsResponse, Self::Error> {
let s = std::str::from_utf8(bytes)?;
Ok(serde_json::from_str(s)?)
}
fn usage<'a>(&self, response: &'a ChatCompletionsResponse) -> Option<&'a Self::Usage> {
Some(&response.usage)
}
fn extract_usage_counts(&self, response: &ChatCompletionsResponse) -> Option<(usize, usize, usize)> {
Some((
response.usage.prompt_tokens as usize,
response.usage.completion_tokens as usize,
response.usage.total_tokens as usize,
))
}
}
impl StreamingResponse for OpenAIProvider {
type Error = OpenAIApiError;
type StreamChunk = ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], _provider: &super::super::ProviderId, _mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let s = std::str::from_utf8(bytes)?;
Ok(OpenAIStreamingResponse::new(s.to_string()))
}
}
// ============================================================================
// Trait Implementations for OpenAI Types (Keep for TokenUsage only)
// ============================================================================
impl TokenUsage for Usage {
fn completion_tokens(&self) -> usize {
self.completion_tokens as usize
}
fn prompt_tokens(&self) -> usize {
self.prompt_tokens as usize
}
fn total_tokens(&self) -> usize {
self.total_tokens as usize
}
}
impl StreamChunk for ChatCompletionsStreamResponse {
type Usage = Usage;
fn usage(&self) -> Option<&Self::Usage> {
self.usage.as_ref()
}
}
impl StreamingResponse for OpenAIStreamingResponse {
type Error = OpenAIApiError;
type StreamChunk = ChatCompletionsStreamResponse;
type StreamingIter = OpenAIStreamingResponse;
fn try_from_bytes(&self, bytes: &[u8], _provider: &super::super::ProviderId, _mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error> {
let s = std::str::from_utf8(bytes)?;
Ok(OpenAIStreamingResponse::new(s.to_string()))
}
}

240
crates/hermesllm/src/providers/traits.rs
View file

@ -4,6 +4,7 @@
//! handling of LLM requests and responses in the gateway. //! handling of LLM requests and responses in the gateway.
use std::error::Error; use std::error::Error;
use std::fmt;
/// Conversion mode for provider requests/responses /// Conversion mode for provider requests/responses
#[derive(Debug, Clone, Copy)] #[derive(Debug, Clone, Copy)]
@ -14,30 +15,41 @@ pub enum ConversionMode {
Passthrough, Passthrough,
} }
/// Trait for provider-specific request types /// Error types for provider operations
pub trait ProviderRequest { #[derive(Debug)]
type Error: Error + Send + Sync + 'static; pub struct ProviderRequestError {
pub message: String,
pub source: Option<Box<dyn Error + Send + Sync>>,
}
/// Parse request from raw bytes #[derive(Debug)]
fn try_from_bytes(&self, bytes: &[u8]) -> Result<crate::apis::openai::ChatCompletionsRequest, Self::Error>; pub struct ProviderResponseError {
pub message: String,
pub source: Option<Box<dyn Error + Send + Sync>>,
}
/// Convert to provider-specific format impl fmt::Display for ProviderRequestError {
fn to_provider_bytes(&self, request: &crate::apis::openai::ChatCompletionsRequest, provider: super::ProviderId, mode: ConversionMode) -> Result<Vec<u8>, Self::Error>; fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "Provider request error: {}", self.message)
}
}
/// Extract the model name from the request impl fmt::Display for ProviderResponseError {
fn extract_model<'a>(&self, request: &'a crate::apis::openai::ChatCompletionsRequest) -> &'a str; fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "Provider response error: {}", self.message)
}
}
/// Check if this is a streaming request impl Error for ProviderRequestError {
fn is_streaming(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> bool; fn source(&self) -> Option<&(dyn Error + 'static)> {
self.source.as_ref().map(|e| e.as_ref() as &(dyn Error + 'static))
}
}
/// Set streaming options (e.g., include_usage) impl Error for ProviderResponseError {
fn set_streaming_options(&self, request: &mut crate::apis::openai::ChatCompletionsRequest); fn source(&self) -> Option<&(dyn Error + 'static)> {
self.source.as_ref().map(|e| e.as_ref() as &(dyn Error + 'static))
/// Extract text content from messages for token counting }
fn extract_messages_text(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> String;
/// Extract the user message for tracing/logging purposes
fn extract_user_message(&self, request: &crate::apis::openai::ChatCompletionsRequest) -> Option<String>;
} }
/// Trait for token usage information /// Trait for token usage information
@ -47,46 +59,178 @@ pub trait TokenUsage {
fn total_tokens(&self) -> usize; fn total_tokens(&self) -> usize;
} }
/// Trait for provider-specific request types
pub trait ProviderRequest: Send + Sync {
/// Extract the model name from the request
fn model(&self) -> &str;
/// Check if this is a streaming request
fn is_streaming(&self) -> bool;
/// Set streaming options (e.g., include_usage)
fn set_streaming_options(&mut self);
/// Extract text content from messages for token counting
fn extract_messages_text(&self) -> String;
/// Extract the user message for tracing/logging purposes
fn extract_user_message(&self) -> Option<String>;
/// Convert to provider-specific format
fn to_provider_bytes(&self, mode: ConversionMode) -> Result<Vec<u8>, ProviderRequestError>;
}
/// Trait for provider-specific response types /// Trait for provider-specific response types
pub trait ProviderResponse { pub trait ProviderResponse: Send + Sync {
type Error: Error + Send + Sync + 'static; /// Get usage information if available - returns dynamic trait object
type Usage: TokenUsage; fn usage(&self) -> Option<&dyn TokenUsage>;
/// Parse response from raw bytes
fn try_from_bytes(&self, bytes: &[u8], provider: &super::ProviderId, mode: ConversionMode) -> Result<crate::apis::openai::ChatCompletionsResponse, Self::Error>;
/// Get usage information if available
fn usage<'a>(&self, response: &'a crate::apis::openai::ChatCompletionsResponse) -> Option<&'a Self::Usage>;
/// Extract token counts for metrics /// Extract token counts for metrics
fn extract_usage_counts(&self, response: &crate::apis::openai::ChatCompletionsResponse) -> Option<(usize, usize, usize)> { fn extract_usage_counts(&self) -> Option<(usize, usize, usize)> {
self.usage(response).map(|u| (u.prompt_tokens(), u.completion_tokens(), u.total_tokens())) self.usage().map(|u| (u.prompt_tokens(), u.completion_tokens(), u.total_tokens()))
} }
} }
/// Trait for streaming response chunks /// Trait for provider-specific streaming response types
pub trait StreamChunk { pub trait ProviderStreamResponse: Send + Sync {
type Usage: TokenUsage; /// Get the content delta for this chunk
fn content_delta(&self) -> Option<&str>;
/// Get usage information if available /// Check if this is the final chunk in the stream
fn usage(&self) -> Option<&Self::Usage>; fn is_final(&self) -> bool;
/// Get role information if available
fn role(&self) -> Option<&str>;
} }
/// Trait for streaming response iterators /// Trait for streaming response iterators
pub trait StreamingResponse { ///
type Error: Error + Send + Sync + 'static; /// This trait ensures that implementing types are iterators that yield
type StreamChunk: StreamChunk; /// ProviderStreamResponse results.
type StreamingIter: Iterator<Item = Result<Self::StreamChunk, Self::Error>>; pub trait ProviderStreamResponseIter: Iterator<Item = Result<Box<dyn ProviderStreamResponse>, Box<dyn std::error::Error + Send + Sync>>> + Send + Sync {
// No additional methods needed - just the Iterator constraint with proper bounds
/// Parse streaming response from raw bytes
fn try_from_bytes(&self, bytes: &[u8], provider: &super::ProviderId, mode: ConversionMode) -> Result<Self::StreamingIter, Self::Error>;
} }
/// Main provider interface trait - simplified to only essential methods // ============================================================================
pub trait ProviderInterface: ProviderRequest + ProviderResponse + StreamingResponse { // PROVIDER FUNCTIONS - NO TRAITS, JUST PARAMETERIZED CONVERSION
/// Check if this provider has a compatible API with the client request // ============================================================================
fn has_compatible_api(&self, api_path: &str) -> bool; //
// ARCHITECTURAL DECISION: Function-based Provider API
//
// We chose this function-based approach over the original ProviderInterface trait
// for several critical reasons:
//
// 1. TRAIT OBJECT LIMITATION:
// - The original ProviderInterface had associated types (Request, Response, etc.)
// - Traits with associated types cannot be used as trait objects (Box<dyn ProviderInterface>)
// - This prevented dynamic provider selection at runtime based on request headers
// - Error: "the trait `ProviderInterface` cannot be made into an object"
//
// 2. DYNAMIC PROVIDER SELECTION REQUIREMENT:
// - The gateway needs to select providers dynamically based on incoming headers
// - Cannot know provider type at compile time - must dispatch at runtime
// - Need ability to return generic trait objects that work polymorphically
//
// 3. WRAPPER TYPE ELIMINATION:
// - Original design required wrapper types like OpenAIRequestWrapper, OpenAIResponseWrapper
// - User wanted to implement traits directly on concrete types (ChatCompletionsRequest, etc.)
// - Function-based approach allows direct trait implementations without wrappers
//
// 4. PARAMETERIZED CONVERSION PATTERN:
// - Follows existing codebase pattern: TryFrom<(&[u8], &ProviderId)>
// - Enables runtime provider selection while maintaining type safety
// - Single implementation can handle multiple OpenAI-compatible providers
//
// 5. TYPE ERASURE FOR GENERIC INTERFACE:
// - Functions return Box<dyn ProviderRequest/Response> - works as trait objects
// - stream_context.rs can work with generic interfaces without knowing concrete types
// - Maintains polymorphism while enabling dynamic dispatch
// ============================================================================
/// Get supported API endpoints for this provider use crate::ProviderId;
fn supported_apis(&self) -> Vec<&'static str>;
/// Parse request from bytes using provider ID - returns generic ProviderRequest trait object
pub fn try_request_from_bytes(bytes: &[u8], provider_id: &ProviderId) -> Result<Box<dyn ProviderRequest>, ProviderRequestError> {
match provider_id {
// All these providers currently use OpenAI-compatible chat completions API
// In the future, we can add provider-specific handling in separate match arms
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek
| ProviderId::Arch | ProviderId::Gemini | ProviderId::Claude | ProviderId::GitHub => {
let request = crate::apis::openai::ChatCompletionsRequest::try_from((bytes, provider_id))
.map_err(|e| ProviderRequestError {
message: format!("Failed to parse request: {}", e),
source: Some(Box::new(e)),
})?;
// Return as trait object - this enables polymorphic usage
// ChatCompletionsRequest implements ProviderRequest directly (no wrapper needed)
Ok(Box::new(request) as Box<dyn ProviderRequest>)
}
}
}
/// Parse response from bytes using provider ID - returns generic ProviderResponse trait object
pub fn try_response_from_bytes(bytes: &[u8], provider_id: &ProviderId, _mode: ConversionMode) -> Result<Box<dyn ProviderResponse>, ProviderResponseError> {
match provider_id {
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek
| ProviderId::Arch | ProviderId::Gemini | ProviderId::Claude | ProviderId::GitHub => {
// Parameterized conversion allows provider-specific response parsing
let response = crate::apis::openai::ChatCompletionsResponse::try_from((bytes, provider_id))
.map_err(|e| ProviderResponseError {
message: format!("Failed to parse response: {}", e),
source: Some(Box::new(e)),
})?;
// ChatCompletionsResponse implements ProviderResponse directly - no wrapper needed!
Ok(Box::new(response) as Box<dyn ProviderResponse>)
}
}
}
/// Create streaming response using provider ID - returns clean ProviderStreamResponseIter trait object
///
/// This function returns a ProviderStreamResponseIter that's just an iterator,
/// eliminating the complex nested Result<Box<dyn Iterator<...>>> type completely.
pub fn try_streaming_from_bytes(bytes: &[u8], provider_id: &ProviderId, _mode: ConversionMode) -> Result<Box<dyn ProviderStreamResponseIter>, Box<dyn std::error::Error + Send + Sync>> {
match provider_id {
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek
| ProviderId::Arch | ProviderId::Gemini | ProviderId::Claude | ProviderId::GitHub => {
// Parse SSE (Server-Sent Events) streaming data
let s = std::str::from_utf8(bytes)?;
let lines: Vec<String> = s.lines().map(|line| line.to_string()).collect();
let iter = crate::apis::openai::SseChatCompletionIter::new(lines.into_iter());
// Return the iterator directly - it implements ProviderStreamResponseIter
Ok(Box::new(iter))
}
}
}
/// Check if provider has compatible API
///
/// Replaces the old ProviderInterface::has_compatible_api method.
/// This function enables runtime API compatibility checking without needing a provider instance.
pub fn has_compatible_api(provider_id: &ProviderId, api_path: &str) -> bool {
match provider_id {
// Currently all these providers support OpenAI chat completions API
// Future providers with different APIs will get their own match arms
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek
| ProviderId::Arch | ProviderId::Gemini | ProviderId::Claude | ProviderId::GitHub => {
api_path == "/v1/chat/completions"
}
}
}
/// Get supported APIs for provider
///
/// Replaces the old ProviderInterface::supported_apis method.
/// Returns a static list of supported API endpoints for the given provider.
pub fn supported_apis(provider_id: &ProviderId) -> Vec<&'static str> {
match provider_id {
ProviderId::OpenAI | ProviderId::Groq | ProviderId::Mistral | ProviderId::Deepseek
| ProviderId::Arch | ProviderId::Gemini | ProviderId::Claude | ProviderId::GitHub => {
vec!["/v1/chat/completions"]
}
}
} }

135
crates/llm_gateway/src/stream_context.rs
View file

@ -10,10 +10,10 @@ use common::ratelimit::Header;
use common::stats::{IncrementingMetric, RecordingMetric}; use common::stats::{IncrementingMetric, RecordingMetric};
use common::tracing::{Event, Span, TraceData, Traceparent}; use common::tracing::{Event, Span, TraceData, Traceparent};
use common::{ratelimit, routing, tokenizer}; use common::{ratelimit, routing, tokenizer};
use hermesllm::providers::traits::{ use hermesllm::{
ProviderRequest, ProviderResponse, StreamChunk, StreamingResponse, TokenUsage, try_request_from_bytes, try_response_from_bytes, try_streaming_from_bytes, ConversionMode,
ProviderId,
}; };
use hermesllm::{ConversionMode, Provider, ProviderId};
use http::StatusCode; use http::StatusCode;
use log::{debug, info, warn}; use log::{debug, info, warn};
use proxy_wasm::hostcalls::get_current_time; use proxy_wasm::hostcalls::get_current_time;
@ -79,8 +79,8 @@ impl StreamContext {
.expect("the provider should be set when asked for it") .expect("the provider should be set when asked for it")
} }
fn get_provider(&self) -> Provider { fn get_provider_id(&self) -> ProviderId {
self.llm_provider().create_provider() self.llm_provider().to_provider_id()
} }
fn select_llm_provider(&mut self) { fn select_llm_provider(&mut self) {
@ -298,9 +298,9 @@ impl HttpContext for StreamContext {
} }
}; };
let provider = self.get_provider(); let provider_id = self.get_provider_id();
let mut deserialized_body = match ProviderRequest::try_from_bytes(&provider, &body_bytes) { let mut deserialized_body = match try_request_from_bytes(&body_bytes, &provider_id) {
Ok(deserialized) => deserialized, Ok(deserialized) => deserialized,
Err(e) => { Err(e) => {
debug!( debug!(
@ -329,10 +329,10 @@ impl HttpContext for StreamContext {
}; };
// Use the provider interface methods for cleaner interaction // Use the provider interface methods for cleaner interaction
let model_requested = provider.extract_model(&deserialized_body).to_string(); // Convert to owned string let model_requested = deserialized_body.model().to_string(); // Convert to owned string
// Extract user message for tracing // Extract user message for tracing
self.user_message = provider.extract_user_message(&deserialized_body); self.user_message = deserialized_body.extract_user_message();
info!( info!(
"on_http_request_body: provider: {}, model requested (in body): {}, model selected: {}", "on_http_request_body: provider: {}, model requested (in body): {}, model selected: {}",
@ -342,15 +342,15 @@ impl HttpContext for StreamContext {
); );
// Use provider interface for streaming detection and setup // Use provider interface for streaming detection and setup
self.streaming_response = provider.is_streaming(&deserialized_body); self.streaming_response = deserialized_body.is_streaming();
// Set streaming options if needed // Set streaming options if needed
if self.streaming_response { if self.streaming_response {
provider.set_streaming_options(&mut deserialized_body); deserialized_body.set_streaming_options();
} }
// Use provider interface for text extraction (after potential mutation) // Use provider interface for text extraction (after potential mutation)
let input_tokens_str = provider.extract_messages_text(&deserialized_body); let input_tokens_str = deserialized_body.extract_messages_text();
// enforce ratelimits on ingress // enforce ratelimits on ingress
if let Err(e) = self.enforce_ratelimits(&model_requested, input_tokens_str.as_str()) { if let Err(e) = self.enforce_ratelimits(&model_requested, input_tokens_str.as_str()) {
self.send_server_error( self.send_server_error(
@ -365,21 +365,18 @@ impl HttpContext for StreamContext {
let _hermes_llm_provider_id = ProviderId::from(llm_provider_str.as_str()); let _hermes_llm_provider_id = ProviderId::from(llm_provider_str.as_str());
// Convert chat completion request to llm provider specific request using provider interface // Convert chat completion request to llm provider specific request using provider interface
let deserialized_body_bytes = match provider.to_provider_bytes( let deserialized_body_bytes =
&deserialized_body, match deserialized_body.to_provider_bytes(ConversionMode::Compatible) {
provider.id(), Ok(bytes) => bytes,
ConversionMode::Compatible, Err(e) => {
) { warn!("Failed to serialize request body: {}", e);
Ok(bytes) => bytes, self.send_server_error(
Err(e) => { ServerError::LogicError(format!("Request serialization error: {}", e)),
warn!("Failed to serialize request body: {}", e); Some(StatusCode::BAD_REQUEST),
self.send_server_error( );
ServerError::LogicError(format!("Request serialization error: {}", e)), return Action::Pause;
Some(StatusCode::BAD_REQUEST), }
); };
return Action::Pause;
}
};
self.set_http_request_body(0, body_size, &deserialized_body_bytes); self.set_http_request_body(0, body_size, &deserialized_body_bytes);
@ -550,16 +547,9 @@ impl HttpContext for StreamContext {
// Parse streaming response using OpenAI-compatible format // Parse streaming response using OpenAI-compatible format
// Since all providers use OpenAI-compatible streaming format // Since all providers use OpenAI-compatible streaming format
let provider = self.get_provider(); let provider_id = self.get_provider_id();
let provider_id =
ProviderId::from(self.llm_provider().provider_interface.to_string().as_str());
match StreamingResponse::try_from_bytes( match try_streaming_from_bytes(&body, &provider_id, ConversionMode::Compatible) {
&provider,
&body,
&provider_id,
ConversionMode::Compatible,
) {
Ok(mut streaming_response) => { Ok(mut streaming_response) => {
// Process each streaming chunk // Process each streaming chunk
while let Some(chunk_result) = streaming_response.next() { while let Some(chunk_result) = streaming_response.next() {
@ -587,14 +577,20 @@ impl HttpContext for StreamContext {
} }
} }
// Extract usage information if available // For streaming responses, we handle token counting differently
if let Some(usage) = chunk.usage() { // The ProviderStreamResponse trait provides content_delta, is_final, and role
let completion_tokens = usage.completion_tokens(); // Token counting for streaming responses typically happens with final usage chunk
self.response_tokens += completion_tokens; if chunk.is_final() {
debug!( // For now, we'll implement basic token estimation
"Streaming chunk completion tokens: {}", // In a complete implementation, the final chunk would contain usage information
completion_tokens debug!("Received final streaming chunk");
); }
// For now, estimate tokens from content delta
if let Some(content) = chunk.content_delta() {
// Rough estimation: ~4 characters per token
let estimated_tokens = content.len() / 4;
self.response_tokens += estimated_tokens.max(1);
} }
} }
Err(e) => { Err(e) => {
@ -605,40 +601,37 @@ impl HttpContext for StreamContext {
} }
Err(e) => { Err(e) => {
warn!("Failed to parse streaming response: {}", e); warn!("Failed to parse streaming response: {}", e);
return Action::Continue;
} }
} }
} else { } else {
debug!("non streaming response"); debug!("non streaming response");
let provider = self.get_provider(); let provider_id = self.get_provider_id();
let response = match ProviderResponse::try_from_bytes( let response =
&provider, match try_response_from_bytes(&body, &provider_id, ConversionMode::Compatible) {
&body, Ok(response) => response,
&provider.id(), Err(e) => {
ConversionMode::Compatible, warn!(
) { "could not parse response: {}, body str: {}",
Ok(response) => response, e,
Err(e) => { String::from_utf8_lossy(&body)
warn!( );
"could not parse response: {}, body str: {}", debug!(
e, "on_http_response_body: S[{}], response body: {}",
String::from_utf8_lossy(&body) self.context_id,
); String::from_utf8_lossy(&body)
debug!( );
"on_http_response_body: S[{}], response body: {}", self.send_server_error(
self.context_id, ServerError::LogicError(format!("Response parsing error: {}", e)),
String::from_utf8_lossy(&body) Some(StatusCode::BAD_REQUEST),
); );
self.send_server_error( return Action::Continue;
ServerError::LogicError(format!("Response parsing error: {}", e)), }
Some(StatusCode::BAD_REQUEST), };
);
return Action::Continue;
}
};
// Use provider interface to extract usage information // Use provider interface to extract usage information
if let Some((prompt_tokens, completion_tokens, total_tokens)) = if let Some((prompt_tokens, completion_tokens, total_tokens)) =
provider.extract_usage_counts(&response) response.extract_usage_counts()
{ {
debug!( debug!(
"Response usage: prompt={}, completion={}, total={}", "Response usage: prompt={}, completion={}, total={}",